DEV Community: Margaret Q. Ombere

RAGS

Margaret Q. Ombere — Mon, 15 Sep 2025 02:50:15 +0000

RAG FOR DUMMIES

What is RAG?

RAG means Retrieval-Augmented Generation.

That’s just a fancy way of saying: “AI that looks things up before answering.”

Without RAG: The AI only answers from what it remembers (its training).
With RAG: The AI can check extra documents or a database, then give a better answer.

Example:

It’s like asking a friend a question:

If they answer from memory, they might forget details.
If they check a book first, the answer will be fresher and more correct.

Why is RAG Useful?

Up-to-date answers – no old information stuck in memory.
Customizable – you can connect it to your own files, notes, or company data.
More accurate – reduces mistakes where AI “makes things up.”

In Short

RAG = AI Memory + Library Access

It makes AI smarter, more reliable, and more useful in real life.

UNSUPERVISED LEARNING

Margaret Q. Ombere — Sun, 31 Aug 2025 15:38:20 +0000

https://medium.com/@maggyombere/unsupervised-learning-288837c8486d

classification in supervised learning

Margaret Q. Ombere — Sun, 24 Aug 2025 18:09:26 +0000

Classification in Supervised Learning:

Supervised Learning in Simple Terms.

Supervised learning is like being taught by example. You get a question, and the teacher gives you the answer. After enough practice, you can solve similar questions on your own.

In machine learning, this means training a model with inputs (features) and their correct outputs (labels), so it can predict hidden cases.

How Classification Fits In

Classification is about chosing between categories. Think of it as answering yes or no, or sometimes choosing between multiple options.

For example:

Is this review positive or negative?
Is this animal a cat, dog, or bird?

The model learns patterns from data and then applies them to new cases.

## Algorithms That Power Classification
Some popular tools include:

Logistic Regression: Quick and effective for simple cases
Decision Trees: Human-friendly and visual
KNN: Straightforward, compares with nearby data points

## My Reflections
I’ve found classification to be both exciting and humbling. At first, it felt easy — “just label things.” But the deeper I went, the more I realized that data quality, feature engineering, and balance matter more than the algorithm itself.

Personally, I love how classification gives immediate, practical results and the outcomes are tangible.

## The Roadblocks I Faced

Imbalanced classes where one label is too rare
Overfitting, especially in complex models
Choosing the right features to feed the model
Making sense of ** algorithms**

API probability of a win

Margaret Q. Ombere — Wed, 30 Jul 2025 21:42:04 +0000

Team Win Probabilities – Season Overview

This summary shows how 20 football teams performed based on their matches played, matches won, and win probability.

🔝 Top Teams

Team	Wins	Win Probability
Liverpool FC	25	0.658
Man City FC	21	0.553
Arsenal FC	20	0.526
Chelsea FC	20	0.526
Newcastle United FC	20	0.526

⚖️ Mid Performers

Team	Win Probability
Aston Villa FC	0.500
Nottingham Forest FC	0.500
Brighton & Hove Albion	0.421
Brentford FC	0.421
Fulham FC	0.395

🔻 Lower Teams

Team	Win Probability
Man United FC	0.289
West Ham United FC	0.289
Ipswich Town FC	0.105
Southampton FC	0.053

Conclusion:

Liverpool led the season in win probability, while Southampton struggled the most. This data highlights clear performance gaps across teams.

Measures of Central Tendency in Data Science

Margaret Q. Ombere — Wed, 30 Jul 2025 21:20:55 +0000

Measures of Central Tendency in Data Science

What Are They?

Measures of central tendency show the center of a dataset. The main ones are:

Mean – average value
Median – middle value
Mode – most frequent value

Why They Matter

Quick summary of data
Spot outliers
Help in model selection
Support data-driven decisions

Example

For 10, 15, 15, 20, 100:

Mean = 32
Median = 15
Mode = 15

The mean is pulled up by the outlier (100), but median and mode reflect the true center.

In summary

These measures simplify complex data and are essential for analysis in data science.

Understanding the relationshipi power Bi and its importance

Margaret Q. Ombere — Wed, 30 Jul 2025 21:07:26 +0000

Power BI Relationships: What They Are & Why They Matter

In Power BI, relationships connect tables using a shared column (like ProductID) so your data model can work properly. Without them, your visuals may be wrong or incomplete.

📘 What’s a Relationship?

A relationship tells Power BI how two tables are linked. For example:

Sales.ProductID connects to Products.ProductID
This lets Power BI know which product each sale refers to

🔢 Types of Relationships

One-to-Many (1:*) – Most common (e.g., Products → Sales)
Many-to-Many (:) – Needs careful handling or a bridge table
One-to-One (1:1) – Rare but useful in master-detail scenarios

✅ Why It Matters

Accurate visuals: Relationships ensure correct totals and filtering
Better models: Cleaner structure with separate tables
Faster reports: Power BI performs better with proper relationships
DAX support: Functions like RELATED() depend on them

NB:

Makes sure one table has unique values
Avoid bi-directional filters unless needed
Use a star schema (fact table at center, dimension tables around)

📌 Strong relationships equals(=) strong reports. Takes a few minutes to review them before you build!

powerbi #lux #datamodeling

How excel is used in real world data analysis

Margaret Q. Ombere — Tue, 10 Jun 2025 17:28:03 +0000

TOPIC: HOW EXCEL IS USED IN REAL-WORLD DATA ANALYSIS
Microsoft Excel is one of the most widely used tools for data analysis across industries. It goes beyond simple spreadsheets to offer powerful features for organizing, analyzing, and visualizing data in meaningful ways.
WAYS EXCEL IS USED
Data Entry and storage/ Data Organization and Cleaning
In real-world data analysis, the first step is often cleaning and organizing messy data. Excel allows users to import data from various sources and clean it using tools like Remove Duplicates, Text to Columns, and Data Validation. This makes it easier to prepare data for analysis.
Statistical Analysis/formulas and calculations
Excel offers a variety of built-in functions for basic statistical calculations, such as AVERAGE, MEDIAN, STDEV, and more. With the Data Analysis Toolpak add-in, users can perform more advanced tasks like regression analysis, t-tests, and histograms — all within Excel.
Charts and visualization/Pivot Tables and Dashboards
Pivot tables allow users to summarize large datasets with just a few clicks. They help answer key questions by sorting, counting, and totaling data. When combined with slicers, charts, and formulas, users can build interactive dashboards to share insights with others.
EXCEL FEATURES AND FORMULAS

SUM =SUM (A1:A10) Adds all numbers in a range (e.g., from A1 to A10 etc ).
AVERAGE =AVERAGE (B2:B12) Calculates the mean (average) of the values in a range.
IF =IF(C2>50, "Pass", "Fail") Checks a condition and returns one value if true, another if false. Conclusion Learning Excel has significantly transformed my understanding of data. Instead of viewing data as isolated numbers, I now see it as structured information that can reveal trends, patterns, and insights. Excel's analytical tools have helped me approach data more critically and interpret it with greater clarity and purpose.